Low latency via redundancy

TL;DR

This paper explores how redundancy can be used to significantly reduce both average and tail latency in networked systems by initiating duplicate operations and using the fastest result, effectively converting extra capacity into lower latency.

Contribution

It provides a theoretical analysis of the tradeoffs involved in task replication for latency reduction and empirically demonstrates its effectiveness in real-world systems.

Findings

Redundancy can reduce mean and tail latency in DNS, databases, and networks.

Replicating all tasks can be beneficial under certain system utilization conditions.

Empirical results show significant latency improvements with redundancy.

Abstract

Low latency is critical for interactive networked applications. But while we know how to scale systems to increase capacity, reducing latency --- especially the tail of the latency distribution --- can be much more difficult. In this paper, we argue that the use of redundancy is an effective way to convert extra capacity into reduced latency. By initiating redundant operations across diverse resources and using the first result which completes, redundancy improves a system's latency even under exceptional conditions. We study the tradeoff with added system utilization, characterizing the situations in which replicating all tasks reduces mean latency. We then demonstrate empirically that replicating all operations can result in significant mean and tail latency reduction in real-world systems including DNS queries, database servers, and packet forwarding within networks.